Will the real Anthropic please stand up

But Anthropic has spent years getting better at telling stories. In fact, its public messages are a master class in shaping how people see the world.

Earlier this month, Jack Clark, one of the company’s founders and its head of policy, gave a speech at an International Monetary Fund (IMF) meeting in Bangkok. He told the economists, bankers, and lawmakers there that the singularity had already begun.

The idea of the singularity was introduced by science fiction writer and futurist Vernor Vinge about 40 years ago. It has also been debated for a long time in more technical corners of the internet. The singularity is supposed to be the point when technology becomes so advanced that it surpasses the abilities of the people who created it.

According to this idea, humans are left behind. They can no longer compete with the results of their own creations, or even understand them.

And that fictional world many of us have imagined since childhood? Apparently, we are already living in it. “I believe that this year we have gained convincing evidence that it has now begun,” Clark told his audience.

The company tells a powerful, and often frightening, story about a future it presents as unavoidable, even though that future has not arrived yet. It often borrows ideas from familiar science fiction stories. It describes its technology as mysterious and dangerous, and says it is the only company responsible enough to control it. What investor would not want to be part of that story?

Here is another example. Last year, Chris Olah, another Anthropic cofounder, was invited to speak at the Vatican. He told Pope Leo XIV and other religious scholars and officials what his company was discovering about its technology. “I will be honest: We keep finding things that are mysterious, even unsettling,” he said. “We find evidence of introspection. We find internal states that functionally mirror joy, satisfaction, fear, grief, and unease.”

Many researchers would strongly disagree with the suggestion that Anthropic’s software has inner experiences like those of humans. But when so little about AI seems certain, people grab onto statements from Anthropic and other leading AI companies and hold on to them as if they were pieces of a sinking ship.

I am not saying that these companies are necessarily wrong about any of this. That is a question for another day. But when nobody knows what will happen next, companies like Anthropic get to fill the gap with their own explanations.

It is worth stopping to ask: What is Anthropic? It might sound like an easy question on a family quiz show, but the answer is much harder than you might expect.

I have been thinking about this question since reading an exchange on X last week. The discussion was started by Nathan Calvin, the chief strategy officer at Encode AI, a Washington, DC-based organization that supports regulation. Calvin was unhappy with legislation Anthropic had proposed in a draft bill that it secretly shared with US lawmakers. The draft was later leaked.

Calvin said Anthropic wants people to see it as the good guy. But why should anyone trust it? “What is Anthropic?” he asked.

Here is one possible answer: Anthropic is a company whose initial public offering (IPO) values it at more than $1.5 trillion. That would make its cofounders, including Clark and Olah, extremely wealthy.

But Anthropic has always said it is more than a business. As Calvin pointed out, it could instead be seen as an AI safety organization that aims to develop powerful technology that is reliable and almost free of risk. Or it could be seen as a group of people determined to create technology with abilities beyond those of humans before anyone else does. They might believe they are the most responsible people to build it, or they might be driven by their own egos.

In fact, Anthropic is all of these things. CEO Dario Amodei made this point in a reply to Calvin’s post: “Our strategy has always been what I’ve called the ‘race to the top’—by showing how it’s possible to build the technology responsibly, we can raise the bar for the entire industry. The forces that drive the race to the top are hugely more effective if the company practicing them is at the frontier; that’s one lesson I’ve learned repeatedly. And looking at that strategy as a whole, I’m proud of our track record and stand by it.”

Amodei makes it sound as if he has everything under control, but playing several roles at once is difficult. For example, an organization focused on AI safety and a giant company driven by profits have very different priorities. Amodei admits in his post that he has made mistakes.

In the end, the stories Anthropic tells itself should help us understand the stories it tells the public. They should help us decide how much to trust the company to regulate itself. They should also help us decide whether we should require outside safety experts to test its models, with access set by our rules rather than the company’s. And they should help us decide whether to give Anthropic the benefit of the doubt when legislation that seems to serve its own interests comes to light.

But I think these stories reveal a many-headed creature. Anthropic means many different things to different people, and I am not sure that makes it more trustworthy. In many ways, a traditional technology company, without the science fiction role-playing and strange claims, would be much easier to understand.